J. Appl. Entomol. 131(9-10), 625–629 (2007) doi: 10.1111/j.1439-0418.2007.01223.x

2007 The Authors Journal compilation 2007 Blackwell Verlag, Berlin

Predatory abilities favour the success of the invasive Pheidole megacephala in an introduced area

A. Dejean1, M. Kenne2 and C. S. Moreau3 1CNRS-Guyane (UPS 2561 and UMR-CNRS 5174), Cayenne, France; 2De´partement de Biologie des Organismes Animaux, Faculte´des Sciences de lÕUniversite´de Douala, Douala, Cameroon; 3Museum of Comparative Zoology, Harvard University, Cambridge, MA, USA

Ms. received: March 8, 2007; accepted: June 19, 2007

Abstract: The invasive African big-headed ant, Pheidole megacephala, is a dominant species in the many areas it has invaded. We examined whether one potential reason for its ecological success might be its predatory efficiency. We compared the density of termite nests in an area of Mexico invaded by P. megacephala with an adjacent area where P. megacephala is not present. We also compared the success of P. megacephala in preying on termites with that of 13 native ant species. We found that termite nest density was significantly lower in areas invaded by P. megacephala (0.33 vs. 1.05 nests per 30 m transect). In field trials, we established that P. megacephala workers were significantly more successful at capturing termite workers from termite nest fragments than even the most successful native ant species, pyramicus. For both P. megacephala and D. pyramicus, single scouts could trigger the mass recruitment of nestmates, but P. megacephala was able to recruit greater numbers of nestmates. Combined with their aggressiveness towards other ant species, their highly efficient predatory capacities help explain the ecological success of P. megacephala and demonstrate how it can be a major threat to invertebrate biodiversity in the areas it invades.

Key words: competition, invasive , Mexico, native ants, predation

1 Introduction terms of abundance and biomass (Wilson 1992). Here we focus on the impact of invasive ants on native Of the approximately 11 500 ant species known termite communities to determine whether they prey (Bolton et al. 2006), about 150 Ôtramp speciesÕ have on native termites or whether termites are unaffected been transported and introduced to many parts of the because they have evolved several means of resistance world through human activity, but only seven have (Ho¨ lldobler and Wilson 1990) that invasive ants are become invasive on a worldwide scale (Holway et al. unable to overcome. 2002). Many of these invasive ants form supercolonies This study, conducted in Mexico, focuses on Phei- due to their intrinsic ability to achieve unicoloniality, dole megacephala (F.), an invasive ant species known which results in the absence of intraspecific territori- to have the ability to eliminate native ants (Heterick ality over extended areas. This feature has been 1997; Hoffmann et al. 1999). Like other invasive ant demonstrated for several of these invasive species species, P. megacephala is thought to be an effective (Tsutui et al. 2003; Fournier et al. 2005). The preval- predator as abundance declines in areas it ence of invasive ant species in areas they have invaded has invaded (Simmonds 1958; Zimmerman 1970; has frequently been associated with their ability to Fowler et al. 1990; Hoffmann et al. 1999). We, there- utilize native food sources, including attending large fore, tested whether the density of arboreal termite numbers of hemipterans for their honeydew (Holway nests was lower in areas P. megacephala has invaded et al. 2002; Le Breton et al. 2005). Moreover, as compared with non-invaded areas, and whether arthropod diversity and abundance has been shown P. megacephala was more successful than native species to decline in invaded areas, invasive ants have been at capturing termites. assumed to be good predators (Holway et al. 2002). As much research has been done on the effect of invasive ants on native arthropod communities, it is surprising, then, to note that no study has ever specifically 2 Materials and Methods addressed the influence these invasive ants have had The present study was conducted along the Caribbean coast on termite abundance, even though in most humid of Quintana Roo, Mexico, around Puerto Morelos (20N, tropics ants and termites constitute the major taxa in 86W) where P. megacephala spread over several kilometres 626 A. Dejean, M. Kenne and C. S. Moreau

(Durou et al. 2002). Along this coast, a strip of land about soldiers and brood from four native species (details provided 450 m wide separates several lagoons from the sea. The in table 1). The termite workers quickly obstructed openings lagoons are connected to the sea by small tidal inlets. The with dirt and secretions, so that some termite individuals coastal strip of land is joined to the inland in several places. were trapped outside the nest fragments and patrolled the Moving from the sea to the inland lagoons, four ecological surface. These termite nest fragments were then installed on zones succeed one another: the beach, dunes with shrubs and either the territory of P. megacephala or on the territory of 13 coconut trees and ⁄ or in places a wide, disturbed lower dune native ant species. For the seven ground-nesting ant species, with sparse vegetation, and finally mangrove all along the the termite nest fragment was placed 5 m away from nest edges of the inland lagoon. entrances. To test the predatory abilities of the three arboreal Native to tropical Africa, P. megacephala has been ant species included in this study, we placed the termite nest unwittingly dispersed by human activity throughout the fragments in a fork of the antÕs host tree more than 2 m from tropics and sub-tropics, becoming one of the most successful the ant nests. Finally, for the three army ant species invasive ant species in these regions (Ho¨ lldobler and Wilson (Ecitoninae), we placed the termite nest fragments about 1990; Hoffmann et al. 1999; Holway et al. 2002; Taylor 0.5 m from a column (but not at the front of the columns in 2006). The worker caste is dimorphic with no intermediary order to reproduce a similar situation as for other ants, with body size between the small minors (approximately 2 mm discovering workers recruiting nestmates). We conducted one long; average 0.35 mg) and the big-headed majors (or test at a time, with their sequence arbitrarily assigned. We soldiers; 3–4 mm long; average 1.65 mg); workers have an verified if the ants (1) attacked, killed and retrieved some of atrophied sting that they use to lay scent trails, but not to the termites patrolling the surface of the tested termite nest subdue prey or competitors (Wilson 2003). Colonies are fragments, (2) if all patrolling termites were captured and (3) found nesting in the ground, in abandoned termite nests or in 2 h after the beginning of the experiment when the ants had the crevices of tree bark and, like most other invasive ants, stopped patrolling the termite nest fragments in most cases, P. megacephala is omnivorous and able to tend many we broke them open to see whether they sheltered termites or hemipteran species (Dejean et al. 2005). not. To assess the impact of the P. megacephala supercolony on In a second series of experiments we tested whether the the abundance of the native arboreal termite Nasutitermes presence alone of chemicals produced by the termites is mexicanus Light, we compared the number of termite nests enough to trigger long-range recruitment in P. megacephala found across the homogeneous mangrove located in an area and Dorymyrmex pyramicus (Roger) scouts (both species occupied by P. megacephala (n = 15) and in a neighbouring were able to capture all termites in the previous test). Prior to area without P. megacephala (n = 20). The plots, 30 m in each experiment, we placed two ÔcleanÕ sheets of A4 paper length, were limited to the first 5 m of mangrove (this part of (21 · 29.7 cm) 5 m from a P. megacephala or D. pyramicus the mangrove is not under water) and separated from each nest entrance (only one test per nest entrance), and counted other by 50 m; we left a 500 m long zone between the the number of workers on these sheets of paper 15 min after monitored invaded area and the control (non-invaded) area. they were discovered by a scout (pre-controls). We then We compared the results using StudentÕs t-test. placed an empty 3–4 cm3 piece of Microcerotermes nest at the In the first series of experiments we began by isolating centre of an ÔexperimentalÕ sheet of paper, while we placed termite nest fragments (about 1 dm3) containing workers, a piece of local dirt of the same volume at the centre of a

Table 1. Termite species mortality tested against foraging native ant species and Pheidole megacephala workers under natural conditions. In each experiment a standardized termite nest fragment (about 1 dm3) was deposited on the territory of a colony, or for Ecitoninae, at about 0.5 m from a column. <2h: ¼ all termites killed in each experiment in less than two hours (no additional workers involved in patrolling the piece of termite nest; no termites recorded within the termite nest fragments). Surface: cases when the ants attacked, killed and retrieved termites patrolling on the termite nest fragments (*: all patrolling termites killed and retrieved) versus total number of cases. First column: P ¼ Ponerinae; Et ¼ Ectatominae; Ec ¼ Ecitoninae (army ants); M ¼ Myrmicinae; D ¼ ; F ¼ Formicinae. **: arboreal-nesting ant species. ***: experiment conducted nocturnally due to the activity rhythm of the species. ÔÔ-ÕÕ: test not conducted

Rhinotermitinae Nasutitermitinae Heterotermes Amitermitinae Nasutitermes mexicanus sp. 1 and sp. 2 Microcerotermes sp. Light

Native ant species <2 h Surface <2 h Surface <2 h Surface

P Anochetus emarginatus (F.) 0 6* ⁄ 6 – – 1 10* ⁄ 10 P Pachycondyla harpax (F.) 0 6* ⁄ 604*⁄ 403⁄ 5 P Pachycondyla stigma (F.) 0 8*8 0 4*4 0 5 ⁄ 5 P Pachycondyla villosa (F.) ** 0 5 ⁄ 5––05⁄ 5 Et Ectatomma ruidum Roger 0 5* ⁄ 504*⁄ 405⁄ 5 Et Ectatomma tuberculatum Olivier 0 5 ⁄ 504⁄ 405⁄ 5 Ec Eciton burchelli (Westwood) 1 5* ⁄ 514*⁄ 405⁄ 5 Ec Eciton hamatum (F.) 0 4 ⁄ 504⁄ 404⁄ 5 Ec Labidus praedator (F. Smith) 2 5* ⁄ 544*⁄ 405⁄ 5 M Solenopsis geminata (F.) 0 8 ⁄ 804⁄ 405⁄ 5 D Dolichoderus bispinosus Olivier** 0 5 ⁄ 6––05⁄ 5 D Dorymyrmex pyramicus (Roger) 4 10* ⁄ 10 3 6* ⁄ 6 1 10* ⁄ 10 F Camponotus atriceps (F.)*** 0 5* ⁄ 604*⁄ 403⁄ 5 Total (%) 7 (8.75) 77 ⁄ 80 (96.2) 8 (12.5) 64 ⁄ 64 (100) 2 (2.7) 70 ⁄ 74 (94.6) M Pheidole megacephala (F.)*** (%) 7 (100) 7* ⁄ 7 (100) 4 (100) 4* ⁄ 4 (100) 9 (100) 9* ⁄ 9 (100)

2007 The Authors Journal compilation 2007 Blackwell Verlag, Berlin, J. Appl. Entomol. 131(9-10), 625–629 (2007) Predation and the success of an invasive ant 627 sheet of paper serving as a control (n = 10 in each case). We (a) b compared the results using the Kruskal–Wallis test followed 20 post hoc Pieces of Microcerotermes by a test (DunnÕs test); the ÔexperimentalÕ results 15 P. termite nests

between P. megacephala and D. pyramicus were compared workers using StudentÕs t-test (and the sequential Bonferroni-correc- 10 ted multiple tests). All statistics were performed using

Number of 5 a GraphPad Prism 4.0 software. aa

Because we observed P. megacephala workers preying on megecephala 0 newly-emerged flies, we placed five garbage cans ⁄ 250-l cylin- Pre-control 1 Pre-control 2 Control Experiment drical tanks (whose bases were pierced with numerous holes about 3 cm in diameter) with thousands of fly brood inside on (b) 20 the territory of a P. megacephala colony for several weeks and then, at night when the ants are the most active, observed their 15 Pieces of Microcerotermes termite nests capacity to capture large numbers of prey. The workers used D. mass recruitment nightly to prey on flies (8 mg on average; workers 10 notably they did not prey on maggots or only on a few). The b

Number of 5 number of captured flies was evaluated during a single night aaa from 54 series of observations along the main colony trails (six pyramics pm am 0 observations per hour between 7:00 and 4:00 .). Pre-control 1 Pre-control 2 Control Experiment

Fig. 1. Testing the influence of termite nestsÕ chemicals 3 Results on the recruitment behavior of Pheidole megacephala and Dorymyrmex pyramicus scouts. Number of recruited The number of Nasutitermes termite nests present in individuals according to the different experiments. ÔÔPre- the first 5-m wide zone of the mangrove was signifi- controlsÕÕ correspond to ÔÔcleanÕÕ sheets of paper, ÔÔcontrol cantly lower in the area occupied by the P. megacep- casesÕÕ to sheets of paper with a piece of dirt of about 3 cm3 hala supercolony than in the control area (0.33 ± 0.13 deposited at the centre of the paper, and ‘‘experimental termite nests per 30 m-long plots; n = 15 vs. cases’’ to sheets of paper with a Microcerotermes nest 1.05 ± 0.15; n = 20; t = 3.34; d.f. = 33; P < 0.01). fragment devoid of occupants of about 3 cm3 (N ¼ 10 in We observed during the first series of experiments each case) deposited at the center of the paper. Global that after discovering the termite nest fragments and comparisions, Kruskal–Wallis tests, P. megacephala: 3 3 without ever coming into contact with the termites, the H 40 ¼ 17.8; P < 0.001; D. Pyramicus:H40 ¼ 18.5; P. megacephala scouts returned to their nests to recruit P < 0.001. Post hoc comparisions, or Dunn’s test: several dozen nestmates. The recruited workers then different letters indicate significant differences rapidly captured and retrieved the termite workers and (P < 0.05). Comparison between the two experimental soldiers patrolling the surface of the termite nest lots, Student’s t-test: t ¼ 7; df ¼ 18; P < 0.01 (after fragments. A few minutes later, several recruited sequential Bonferroni correction). workers dug holes in the termite nest fragments and all the termites were captured (table 1). Most termite During the second series of experiments we observed workers and Nasutitermes soldiers were captured by P. megacephala scouts first antennated the ÔemptyÕ single P. megacephala individuals, rarely two, while the termite nest fragments, and then returned directly to larger soldiers of the other termite genera were their nest to recruit nestmates. Consequently, the cooperatively retrieved by two to four workers. In number of P. megacephala workers present 15 min contrast, among the native ant species only the army after the scouts discovered the sheets of paper was ants Eciton burchelli (Westwood) and Labidus praeda- significantly higher on those with termite nest frag- tor (F. Smith) plus Anochetus emarginatus (F.) and ments than all control cases for which the numbers of D. pyramicus (table 1) raided the termite nest frag- workers were always low (fig. 1a). The same was true ments, rarely killing and retrieving all the termites. for D. pyramicus, the most efficient native species from Using FisherÕs exact test we noted a significant the previous experiment, although it recruited signifi- difference between the most efficient native ant species cantly fewer workers than P. megacephala (fig. 1b). (pooled together) and P. megacephala in their ability to Finally, P. megacephala workers retrieved during raid native termites: Heterotermes (7 of 20cases vs. 7 of each 10-minute series of observations 316 ± 56 7; P < 0.01); Nasutitermes mexicanus (2 of 20 cases vs. newly-emerged flies, resulting in an estimated 9 of 9; P < 0.001); and all termite species pooled (17 of 17 064 ± 3024 flies per night (or about 130 g of fresh 54 cases vs. 20 of 20; P < 0.0001); but not for prey). We also observed during these raids that some Microcerotermes sp. due to low observation numbers workers remained immobile facing away from the (8 of 14 cases vs. 4 of 4; P = 0.16). Nevertheless, all trails, mandibles open, and so served as guards as do the native ants tested attacked, killed and retrieved certain army ants when protecting their columns. some of the termite individuals patrolling the outside of the termite nest fragments (although very few species captured all of them; table 1). Additionally 4 Discussion A. emarginatus workers captured Nasutitermes indi- viduals without triggering aggressiveness from the This study demonstrates that P. megacephala is an termite soldiers. efficient generalist termite predator as illustrated by

2007 The Authors Journal compilation 2007 Blackwell Verlag, Berlin, J. Appl. Entomol. 131(9-10), 625–629 (2007) 628 A. Dejean, M. Kenne and C. S. Moreau the fact that it raided all the native termite nest introduced. As shown, P. megacephala easily captures fragments regardless of the termite species (including termites, which could greatly decrease the termite Nasutitermes that are able to defend territories from populations themselves as well as populations of native dominant ant species in Neotropical countries; native termite specialists. In addition, occasional Dejean et al. 2003). In comparison native ants were termite predators might find it difficult to satisfy far less efficient, even though all the species observed their nitrogen requirements in areas where P. mega- are at least occasional termite predators (concerns cephala have been introduced. Taken together, these termite individuals patrolling the surface of tested results show that P. megacephala is a potential threat termite nest fragments), showing that termites may be to invertebrate diversity in areas where it has been an important protein food source for these species. It introduced due to its ability to raid native with is not surprising that army ants were able to kill and even the most well-developed defences, such as retrieve all the termite individuals in some of the termites. tested termite nest fragments (see Gotwald 1995), but the present study reveals that A. emarginatus and Acknowledgements particularly D. pyramicus can also do the same. Several observations of columns of workers raiding We are grateful to Roy R. Snelling for the identification of arboreal termite nests confirmed that D. pyramicus is the ants, to Andrea Dejean for proofreading early versions of an effective termite predator (A.D. pers. obs.). Con- the manuscript, and to anonymous referees for their cerning our observations of A. emarginatus, it has constructive comments. C.S.M. would like to thank Stefan already been noted that workers are able to patrol P. Cover and Edward O. Wilson for many helpful discussions regarding Pheidole. among Nasutitermes individuals without being attacked by the soldiers; a subsequent analysis of the cuticular hydrocarbons revealed the existence of References a chemical mimicry between these ants and Nasutit- Bolton B, Alpert GD, Ward PS, Naskrecki P, 2006. BoltonÕs ermes soldiers (Dejean 1988). catalogue of ants of the world (C.D) Harvard University The chemicals produced by termites and present on Press, Cambridge. their nests serve as kairomones that triggered the Dejean A, 1988. Les e´comones implique´es dans la pre´dation recruitment of nestmates of both D. pyramicus and chez les fourmis. Ann. Soc. Entomol. Fr. 24, 456. P. megacephala scouts, and, in this way, the individual Dejean A, Durou S, Olmsted I, Snelling RR, Orivel J, 2003. scouts avoid termite defences and remain safe. Nest site selection by ants in a flooded Mexican The raiding survey conducted on flies permitted us mangrove, with special reference to the epiphytic orchid to verify that P. megacephala workers have the ability Myrmecophila christinae. J. Trop. Ecol. 19, 325–341. to capture large numbers of prey (see also Simmonds Dejean A, Le Breton J, Suzzoni JP, Orivel J, Saux-Moreau C, 1958). A similar behaviour could take place unseen in 2005. Influence of interspecific competition on the recruitment behavior and liquid food transport in the termite nest galleries with large P. megacephala colon- tramp ant species Pheidole megacephala. Naturwissens- ies possibly destroying entire termite colonies, while chaften 92, 324–327. most ant species specialized in termite predation only Durou S, Dejean A, Olmsted I, Snelling RR, 2002. Ant capture relatively small parts of colonies, and repeat- diversity in coastal zones of Quintana Roo, Mexico, edly prey on individual termite nests (Ho¨ lldobler and with special reference to army ants. Sociobiology 40, Wilson 1990). 385–402. Because termites are one of the most abundant taxa Fournier D, Estoup A, Orivel J, Foucaud J, Jourdan H, in the humid tropics (Wilson 1992) and most ant Le Breton J, Keller L, 2005. Clonal reproduction by species either opportunistically or specifically prey on males and females in the little fire ant. Nature 435, termites that partly or entirely satisfy their protein 1230–1234. Fowler HG, Bernardi JVE, Delabie JHC, Forti LC, Pereira requirements, P. megacephala may be able to eliminate da Silva V, 1990. Major ant problems of South America. native ants through exploitation competition as it is In: Applied myrmecology: a world perspective. Ed. by a particularly efficient termite predator. Indeed, the Vander Meer RK, Jaffe K, Cedeno A, Westview Press, ability of P. megacephala to displace native ants has Boulder, CO, 3–14. been well documented (Haskins and Haskins 1965; Gotwald WH, 1995. Army ants. The biology of social Lieberburg et al. 1975; Majer 1985; Heterick 1997; predation. Cornell University Press, Ithaca, NY. Hoffmann et al. 1999), but this species has never before Haskins CP, Haskins EF, 1965. Pheidole megacephala and been presented as a termite predator. Rather, P. mega- Iridomyrmex humilis in Bermuda: equilibrium or slow cephala has always been considered to be an exotic, replacement. Ecology 46, 736–740. invasive ant species that has severely reduced native Heterick B, 1997. The interaction between the coastal brown Pheidole megacephala arthropod abundance when introduced both in dis- ant, (Fabricius), and other inver- tebrate fauna of Mt Coottha (Brisbane, Australia). Aust. turbed habitats and in tropical rain forests (Zimmer- J. Ecol. 22, 218–221. man 1970; Hoffmann et al. 2002), and has been show Hoffmann BD, Andersen AN, Hill GJE, 1999. Impact of an to have had a particularly strong, negative impact on introduced ant on native rain forest invertebrates: Phei- populations of houseflies in Fiji (Simmonds 1958) and dole megacephala in monsoonal Australia. Oecologia 120, Collembolans in Queensland (Heterick 1997). 595–60. In conclusion, this is the first study to demonstrate Ho¨ lldobler B, Wilson EO, 1990. The ants. Harvard Univer- that the predatory abilities of P. megacephala favour sity Press, Cambridge. its ecological success in an area where it has been

2007 The Authors Journal compilation 2007 Blackwell Verlag, Berlin, J. Appl. Entomol. 131(9-10), 625–629 (2007) Predation and the success of an invasive ant 629

Holway D, Lach L, Suarez AV, Tsutui ND, Case TJ, 2002. Tsutui ND, Suarez AV, Grosberg RK, 2003. Genetic The causes and consequences of ant invasions. Annu. diversity, asymmetrical aggression, and cooperation in Rev. Ecol. Syst. 33, 181–233. a widespread invasive species. Proc. Natl. Acad. Sci. Le Breton J, Jourdan H, Chazeau J, Orivel J, Dejean A, U.S.A. 100, 1078–1083. 2005. Niche opportunity and ant invasion: the case of Wilson EO, 1992. The diversity of life. Harvard University Wasmannia auropunctata in a New Caledonian rain Press, Cambridge. forest. J. Trop. Ecol. 21, 93–98. Wilson EO, 2003. Pheidole in the New World. A dominant, Lieberburg I, Kranz PM, Seip A, 1975. Bermudian ants hyperdiverse ant genus. Harvard University Press, Cam- revisited. The status and interaction of Pheidole mega- bridge. cephala and Iridomyrmex humilis. Ecology 56, 473–478. Zimmerman EC, 1970. Adaptive radiation in Hawaii with Majer JD, 1985. Recolonisation by ants of rehabilitated special reference to insects. Biotropica 2, 32–38. mineral sand mines on North Stradbroke Island, Queens- land, with particular reference to seed removal. Aust. J. AuthorÕs address: Alain Dejean (corresponding author), Ecol. 10, 31–48. CNRS-Guyane, UPS 2561, 16, avenue Andre´Aron, 97300 Simmonds HW, 1958. The housefly problem in Fiji and Cayenne, France. E-mail: [email protected] Samoa. S. Pac. Comm. Quart. Bull. 8, 29–30. Taylor B, 2006. The ants of Africa [WWW document]. URL http://antbase.org/ants/africa/antcover.htm [accessed on 27 July 2007].

2007 The Authors Journal compilation 2007 Blackwell Verlag, Berlin, J. Appl. Entomol. 131(9-10), 625–629 (2007)